Information ExtractionReferatsthemen

CIS, LMU München

Winter Semester 2019-2020

Prof. Dr. Alexander Fraser, CIS

Information Extraction – Reminder• Vorlesung

• Learn the basics of Information Extraction (IE), Klausur – only on the Vorlesung!

• Seminar

• Deeper understanding of IE topics

• Each student who wants a Schein will have to make a presentation on IE

• New: 18 minutes (LaTeX, PowerPoint, Mac)

• THIS MAY CHANGE AS I MAKE THE SCHEDULE!

• Hausarbeit

• 6 page "Ausarbeitung" (an essay/prose version of the material in the slides), due 3 weeks after the Referat

• Optionally: bonus points from practical exercises

Topics

• Topic will be presented in roughly the same order as the related topics are discussed in the Vorlesung

• Most of the topics require you to do a literature search

• There will usually be one article (or maybe two) which you find is the key source

• For some topics, a suggestion will be made on the slide

• If these sources are not standard peer-reviewed scientific articles, YOU MUST SEND ME AN EMAIL 2 WEEKS BEFORE YOUR REFERAT to ask permission

• I am also open to topic suggestions suggested by you, send me an email

Referat

• Tentatively (MAY CHANGE!):• 18 minutes plus about 12 minutes for discussion

• Start with what the problem is, and why it is interesting to solve it (motivation!)

• It is often useful to present an example and refer to it several times

• Then go into the details

• If appropriate for your topic, do an analysis

• Don't forget to address the disadvantages of the approach as well as the advantages

• Be aware that advantages tend to be what the original authors focused on!

• List references and recommend further reading

• Number your slides (useful in discussion)

• Have a conclusion slide!

• IMPORTANT: if your topic is repeated from a previous year's seminar, explicitly (but briefly) say what was done there and how your presentation is different!

Languages

• If you do the slides in English, then presentation in English (and Hausarbeit in English)

• If you do the slides in German, then presentation in German (and Hausarbeit in German)

• You must specify the presentation language when you specifytopics, I will use this in scheduling the topics

References I

• Please use a standard bibliographic format for your references

• This includes authors, date, title, venue, like this:

• Academic Journal• Alexander Fraser, Helmut Schmid, Richard Farkas, Renjing Wang, Hinrich Schuetze

(2013). Knowledge Sources for Constituent Parsing of German, a Morphologically Rich and Less-Configurational Language. Computational Linguistics, 39(1), pages 57-85.

• Academic Conference• Alexander Fraser, Marion Weller, Aoife Cahill, Fabienne Cap (2012). Modeling Inflection and

Word-Formation in SMT. In Proceedings of the 13th Conference of the European Chapter of the Association for Computational Linguistics (EACL), pages 664-674, Avignon, France, April.

References II

• In the Hausarbeit, use *inline* citations:• "As shown by Fraser et al. (2012), the moon does not consist of cheese"

• "We build upon previous work (Fraser and Marcu 2007; Fraser et al. 2012) by ..."

• Sometimes it is also appropriate to include a page number (and you *must* include a page number for a quote or graphic)

• Please do not use numbered citations like:• "As shown by [1], ..."

• Numbered citations are useful to save space, otherwise quite annoying

References III

• If you use graphics (or quotes) from a research paper, MAKE SURE THESE ARE CITED ON THE *SAME SLIDE* IN YOUR PRESENTATION!• These should be cited in the Hausarbeit in the caption of the graphic

• Please include a page number so I can find the graphic quickly

• Web pages should also use a standard bibliographic format, particularly including the date when they were downloaded

• I am not allowing Wikipedia as a primary source

• I no longer believe that Wikipedia is reliable, for most articles there is simply not enough review (mistakes, PR agencies trying to sell particular ideas anonymously, etc.)

• Wikipedia can be useful for background, but please don't cite Wikipedia pages!

• You also cannot use student work (not peer-reviewed by people with PhDs) as a primary source• If in doubt, email me!

Administravia I

• Please send me an email with your preferences• Starting at 18:00 on *Monday*

• Please say which seminar (weekday) you are in (and your name)

• Specify which language you will present in

• Emails will be processed in the order received

• Emails received before 18:00, even one minute before, will be processed later, this is the only fair way to allocate topics

• Please specify multiple topics (ranked)

• Last topics assigned on Thursday next week, this is the deadline!

9

Administravia II

• You can take a look at topics from the previous year (you should know the password)

• You can look at the seminar web page as I update it, click the refresh button in your browser due to possible cachingproblems

• First seminar topics are already in three weeks

10

Administravia III

• Please check the plugs on your laptop and the projector in advance (i.e., VGA? HDMI?)

• Rehearse the talk so that you know it really ends after 18 minutes. I will cut you off shortly after this time limit!

• PLEASE DO NOT FORGET THE SLIDE NUMBERS!

11

Administravia - IV

• Seminars next week are cancelled, Vorlesung is NOT cancelled

• Tutor: Tobias Eder

• Monday 04.11 and Thursday 07.11 Tobias Eder and I will hold an exercise on manually written rules for extraction in the Seminar• People only in the Vorlesung are also invited

• See the Seminar web page for location

• Bonus points (for the Seminar only)

12

• Questions?

13

Information Extraction

Source

Selection

Tokenization&

Normalization

Named Entity

Recognition

Instance

Extraction

Fact

Extraction

Ontological

Information Extraction

?

05/01/67

1967-05-01

And Beyond!

Person Name Person Type

Elvis Presley musician

Angela Merkel politician

Information Extraction (IE) is the process of extracting

structured information from

unstructured machine-readable documents

... married Elvis on 1967-05-01

Tip of the hat: Suchanek

Relation Entity1 Entity2

Married Elvis Presley Priscilla

Beaulieu

CEO Tim Cook Apple

✓

✓

• Some of the topics must be in English

• Two common pitfalls:

• Please provide the motivation for your topic!

• PLEASE DO NOT FORGET SLIDE NUMBERS!

15

History of IE

• TOPIC: MUC, ACE and *focus on* TAC, particularly TAC KBP2019 Ultra-Fine-Grained Name Tagging task• These workshops worked on Information Extraction, funded by US

but a large variety of people participated

• Discuss problems solved, motivations and techniques

• Survey the literature

• MUST BE IN ENGLISH

16

Named Entity Recognition – Entity Classes

• TOPIC: fine-grained open classes of named entities• Survey the proposed schemes of fine-grained open classes, such as BBN's

classes used for question answering

• Discuss the advantages and disadvantages of the schemes

• Discuss also the difficulty of human annotation – can humans annotate these classes reliably?

• How well do classification systems work with these fine grained classes?

• MUST BE IN ENGLISH

17

Rule-based IE vs. Statistical

• TOPIC: Rule-based IE (dominant in industry) vs. Statistical IE (dominant in academia)• Discuss the academic history of IE

• What is the general view in academia towards rule-based IE?

• How is statistical IE viewed in industry?

• Time allowing: hybrid

• MUST BE IN ENGLISH

18

NER – German

• TOPIC: Named Entity Recognition of Entities in German

• What is different about NER for German than for English?

• Which models work best for German, considering both contemporary andhistorical texts? How do they work and what are they trained on?

• Which set of classes is annotated?

• What sort of domain adaptation techniques work here?

• Paper: Martin Riedl, Sebastian Padó (2018). A Named Entity Recognition Shootout for German. ACL 2018

19

NER – Twitter

• TOPIC: Named Entity Recognition of Entities in Twitter

• There has recently been a lot of interest in annotating Twitter

• Which set of classes is annotated? What is used as supervised training material, how is it adapted from non-Twitter training sets?

• What are the peculiarities of working on 140 character tweets rather than longer articles?

• What sort of domain adaptation techniques work here?

• Paper: G Aguilar, S Maharjan, AP Lopez-Monroy, et al. (2019). A multi-task approach for named entity recognition in social media data. arXiv.

20

Instance Extraction – Coreference

• TOPIC: surveying the literature on Coreference• How do existing pipelines work? What are the differences?

• What gold standard data is available for testing systems?

• What types of coreference are detected?

• How do the models work?

• What sort of results does one get?

• What are the open problems?

• PAPER: please select a 2018 or 2019 paper as your primarysource

21

Event Extraction – Disasters in Social Media

• TOPIC: Extracting Information during a disaster from social media (e.g., Twitter)• What sorts of real-time information extraction can be done using

social media?

• What are the entities detected?

• How is the information aggregated?

• How can the information be used?

• PAPER: please select a 2018 or 2019 paper as yourprimary source

20

Creating Training Data with Weak Supervision for Relation Extraction

• TOPIC: using rules instead of hand-labeling training data for relation extraction

• All machine learning based systems are heavily dependent on large training data

• But domain experts can often write rules effectively that capture important generalizations

• Can we use these rules to augment supervised relation extraction systems?

• Recommended Paper:

• Alexander Ratner, Stephen H. Bach, Henry Ehrenberg, Jason Fries, Sen Wu, Christopher Re (2017). Snorkel: Rapid Training Data Creation with Weak Supervision. VLDB 2017.

• (Viktor Hangya, Simon Rieß, Matthias Huck, Jindrich Libovicky, Dario Stojanovski, Alexandra Chronopoulou)

Choosing a topic

• I will put these slides on the seminar page later today

• Please email me with your choice of topics (FOR ALL TOPICS!), starting at *18:00* Monday• Do not forget to include the presentation language (and your name)

• If you are emailing later, check the seminar web page first to see if the topicis already taken!• The final deadline is Thursday!

• Any questions?

• Thank you for your attention!